Published Online: 9 April 2010
Page Count: 11
Research Group Manager, CREAS–ASCOMETAL Research Center, Hagondange Cedex,
(Received 4 November 2009; accepted 4 January 2010)
Today, the cleanliness assessment of bearing steel is usually performed by using standard metallographic methods such as ASTM E45, DIN 50602, Norme Internationale ISO 4967, ASTM 2283, etc. These methods are based on the estimation of indexes, and they use either reference images given by charts (Plate I-r for use with ASTM E45) or the principle of the extreme values. The obtained indexes contribute to the quality assessment of a heat. As all these methods do not give the same results, they must be compared to determine the following: First, what the most appropriate method for a given case is and, second, what the reliability of each obtained result is. It is nearly impossible to answer these questions on the basis of a set of experimental measurements coming from these methods. Indeed, this approach is inevitably time-consuming and does not offer any guarantee as to the conclusions. The main reason is that no standard sample exists with known cleanliness properties. To solve this problem, we have developed a simulation approach. In this case, the different methods of cleanliness assessment are simulated on virtual samples. The inclusion populations are perfectly known in this kind of sample (number of inclusions per mm3, sizes, positions, etc.). To create them, the characteristic parameters of these populations (size distribution, elongation distribution, etc.) must be precisely obtained through experiments. To carry this out, an automatic system of measurement has been developed using a scanning electronic microscope and an energy dispersive spectrometer system. The model in this simulation approach takes into account the experimental conditions (detection limit, observed area, etc.) and gives numerical results according to the typical chart taken from the standard methods. So, it is possible to compare and to evaluate the reliability of the results from the different methods or to quantify the effects of a parameter of a method on the results. Moreover, it provides the reliability of an experimental result to meet the requirement of a given customer in a more precise way.
Paper ID: JAI102834